Although the U.S. Presidential election has passed, we can expect that visual mis- and dis-information will continue to pull and tear at our social fabric. In the last weeks of the Presidential campaign we saw manipulated video of Joe Biden purportedly greeting the wrong state and altered images of prominent celebrities supposedly wearing “Trump 2020” hats. More sophisticated uses of synthetic media were used to generate a fabricated identity of an alleged security investigator and propel a fraudulent report on Hunter Biden. Although these efforts were eventually debunked, they garnered national attention and were seen by hundreds of thousands of potential voters. It is clear that manipulated media is being weaponized to amplify misleading content and polarize our society.
As more manipulated content liters the online landscape, our trust in everything erodes. A video of President Trump, recorded shortly after his hospitalization due to COVID-19, was met with wild speculation that it was manipulated. Similarly, the video of the killing of George Floyd was claimed by some to be fake and a ploy to incite civil unrest. If anything can be manipulated, then everything can be argued to be fake—and we lose the ability to agree on even the most basic facts. This deceptive phenomenon, known as the “Liar’s Dividend,” has been used to undermine reality in conflict zones like Syria, but also domestically.
Big Tech has a fundamental role to play in countering this dangerous trend. As a significant news source for billions of people around the world, these companies have a responsibility to build apolitical and unbiased mechanisms to help establish a commonly agreed-upon record of visual truth. So far, Google, Facebook, Twitter, and others have failed to do this. It’s time for them to change course and devote the requisite effort, energy, and funds to make truth an equal priority to profit.
The past two decades have seen significant progress in the development of techniques for authenticating visual content. But they are still not accurate or fast enough to contend with the flood of digital content uploaded every day: more than 500 hours to YouTube every minute, more than 14 million images to Facebook every hour, and more than 500 million tweets a day. Despite significant research efforts at all levels of the private sector, academia, and the government, even the most optimistic projections put us years away from being able to reliably and accurately authenticate content at the necessary scale and speed.
There is another way to approach the problem: provenance-based capture, which flips the problem of authentication and asks the camera to authenticate images and videos at the point of recording. This technology—which Truepic pioneered in 2015, but is also being developed and expanded upon by the likes of Serelay, Adobe, Microsoft, and others—is viable and already available. We believe it’s the only scalable long-term approach to pushing back against the erosion of trust in what we see and hear online.
A provenance solution implants a digital signature—think of it as a unique fingerprint—into a photo at its creation. The signature, immutably attached throughout the photo’s life cycle, raises the confidence of the person viewing the content by offering high-integrity information (date, time, location, and pixel-level content) that is mathematically guaranteed to be unmodified. This holds enormous benefit to anyone making a decision based on visual content—whether it’s someone considering an online purchase or the UN Security Council addressing images from a conflict zone.
Until now, provenance-based audio, image, and video capture was only available through apps that smartphone users had to download and remember to use. This hindered the technology’s reach and limited its security. However, a recent engineering breakthrough by Truepic engineers working on Qualcomm’s Snapdragon chipset now allows image authentication technology to be built directly into smartphone hardware. This means that smartphone makers will have the ability to provide secure image and video capture functionality in the cameras that people use daily, empowering billions of people to disseminate authenticated information.
The breakthrough comes at the same time that several initiatives, backed by large technology companies, have helped standardized trusted image and video formats, so that they can be viewed and understood across any online service. In sum, there is now a viable method for anyone in the world to capture and share visual truth. This holds the potential to revolutionize how people communicate.
For this technology to be successful, though, it will need to be integrated into our online experience from the point of capture to distribution and consumption. The third step, consumption, is arguably the most important. It’s time for Big Tech to step up and ensure that social media platforms, web browsers, and other content channels both recognize and display these images with verified provenance. Social media does not need to be the arbiter of truth: It can instead lean on this model to help users make more informed decisions around the media they consume. Both social media and mainstream media can, without the fear of bias, prioritize and promote information embedded with authenticated data. And the U.S. government can regain people’ trust by mandating that all official U.S. government media be recorded with provenance-based capture technology.
Misinformation and the malicious use of synthetic media is, of course, an issue that will not be solved just by technologists. It is a very human problem. This breakthrough, however, represents a significant tool that, in time, can help empower the world to restore a shared sense of reality in digital media. The only question now is whether social media platforms, governments, academics, and others will get behind provenance-based technology and help build out the infrastructure necessary to restore trust, both online and off.
Hany Farid is a Professor in Electrical Engineering and Computer Sciences and School of Information at the University of California, Berkeley, and an Advisor to Truepic.
Jeff McGregor is the CEO of Truepic.